fix #953 Possible reflective cross site scripting in ConstraintViolations

[WarFox]

• ConstraintViolations should not print the invalid value as it may cause reflective-cross-site scripting issue
• Fixed all unit tests that expect the (was {invalidValue}) in the violation message.

[piefel]

Since the format() call is also used for the trace logging in JacksonMessageBodyProvider where printing the invalid value is harmless but helpful, I wonder whether there couldn’t be a second format function that would exhibit the classic behaviour.

[piefel]

Since the format() call is also used for the trace logging in JacksonMessageBodyProvider where printing the invalid value is harmless but helpful, I wonder whether there couldn’t be a second format function that would exhibit the classic behaviour.

[carlo-rtr]

I think we should add a debug log with the invalid value, so it can be debugged on the server side.

I think this would also address @piefel concern without having to maintain a flag or separate methods.

@joschi you commented on #892 which is related, what do you think about this?

[carlo-rtr]

I think we should add a debug log with the invalid value, so it can be debugged on the server side.

I think this would also address @piefel concern without having to maintain a flag or separate methods.

@joschi you commented on #892 which is related, what do you think about this?

[joschi]

I'm all for it. Let's just remove the original, tainted input from the error messages.

[joschi]

I'm all for it. Let's just remove the original, tainted input from the error messages.

[WarFox]

I agree with @carlo-rtr about the debug logs, that could be of help to some projects.

Should we add the debug logs?

[WarFox]

I agree with @carlo-rtr about the debug logs, that could be of help to some projects.

Should we add the debug logs?

[carlo-rtr]

I think so :)

[carlo-rtr]

I think so :)